Project-Level Model Provider
Model Provider configuration is managed at the project level in Robility Flow. A new Model Providers option is available under Global Variables in the Canvas Flow sidebar, allowing each project to configure and manage its own AI providers and language models.
Each project has its own model provider configurations, credentials, and models. These settings are available only within the project and can be used by its AI components and the Flow Assistant.
Key Benefits
- Configure AI providers independently for each project.
- Keep provider configurations isolated between projects.
- Make only project-specific providers available to AI components and the Flow Assistant.
- Use different AI providers, models, and credentials for different projects.
How AI Components Use Model Providers
After a model provider is configured, it becomes available to AI components and the Flow Assistant in the current project.
When configuring an AI component, select the required model provider and language model. During execution, the component uses the selected provider configuration to connect to the AI service.
Important:
- Model Provider configurations are not migrated automatically at the project level. Reconfigure them from the Model Providers menu on the Canvas Flow page.
- Language Model selections are not retained. Update the Language Model manually for the required components.
- Model Provider validation happens during execution. Errors are displayed if the Model Provider is not configured correctly for the project.
- Playground does not validate project-level Model Provider configurations. Configuration issues are identified only during component execution.
Configure a Model Provider
- Open the Canvas Flow page.
- Click the Brain icon in the right panel.
- Select an AI model provider.
- Enter the required details, such as:
- API Key
- Project ID
- Endpoint URL
- Click Save.
The configured provider is then available to AI components and the Flow Assistant in the current project.
Secure Sensitive Values
Store sensitive values, such as API keys and access tokens, in the Vault instead of entering them directly in the configuration.
Reference the Vault entries when configuring the model provider. During execution, Robility Flow securely retrieves the stored values, preventing sensitive information from being exposed to other users.
Best Practices
- Store all API keys, secrets, and credentials in the Vault.
- Reference Vault entries in model provider configurations.
- Do not hardcode sensitive values in workflows or project configurations.
